your friendly ai assistant
Find a file
2025-08-02 21:39:23 +02:00
.gitignore initial commit 2025-07-18 12:53:00 +02:00
config.example.toml image vision (cooks pcs) 2025-08-02 21:39:23 +02:00
lydia.js image vision (cooks pcs) 2025-08-02 21:39:23 +02:00
package.json image vision (cooks pcs) 2025-08-02 21:39:23 +02:00
README.md image vision (cooks pcs) 2025-08-02 21:39:23 +02:00

lydia

your friendly ai assistant. frontend for ollama.

fair warning: lydia is very stupid

Installation

  • git clone repository
  • install ollama from https://ollama.ai/download
  • pull a model from ollama (i recommend gemma3n:e4b for laptops like mine (i7-10750h + rtx 3050ti laptop edition))
  • for image stuff, you'll need a model like llava:7b
  • for webcam features, install fswebcam (Linux) or imagesnap (macOS)
  • copy config.example.toml to config.toml and edit it to have the model you selected, optionally set your name in [user]
  • npm i
  • node index.js
  • enjoy

Configuration

lydia is written to be easily configurable through a toml file which is easier to read and understand than json

Assistant settings

  • name = change the name you want the assistant to use for itself. changes it everywhere, including all placeholder text. default is "lydia"
  • model = the ollama model you want lydia to use. default is "gemma3n:e4b"
  • system_prompt = the system prompt you want lydia to use.
  • assistantface = the default face you want lydia to have before waking up. default is "=w="

Apperance settings

  • facefont = the font you want lydia to use for her face kaomoji. default is "mono9". these are figlet fonts, like ".flf" files. you cannot use TTF or OTF font types. a good preview is here: https://www.askapache.com/online-tools/figlet-ascii/. you must have these fonts installed on your system, or install them depending on your operating system.

User settings

  • name = the name you want lydia to call you. default is "user"

Advanced configuration

(I wouldn't touch this unless you know what you're doing)

  • temperature = the temperature you want lydia to use. basically how random the model is. default is 0.8
  • max_tokens = the max context tokens you want lydia to use. basically how far she can remember. default is 8192

Camera settings

  • width = webcam capture width. default is 1280
  • height = webcam capture height. default is 720
  • quality = webcam capture quality (0-100). default is 100
  • device = specific camera device (false for default, or "/dev/video0", etc.)

Runtime configuration

the prompt can be changed by running l!prompt in the chatbox. this only applies for the current session, if you want a persistent change, you can edit the config file.

Image & Webcam Commands

  • l!image <path> or l!img <path> - send an image file (or just l!image to browse)
  • l!webcam or l!cam - take and send a webcam snapshot
  • press ESC to open menu for image options

Other stuff

by hitting escape you can open the menu, here you can do cool things like:

  • send pictures and take webcam snapshots
  • get help and see all commands
  • hit Q or CTRL+C to quit lydia (but why would you wanna do that anyway?)